1. Field of the Invention
This invention relates to a device and method for modulation and transmission medium, and more particularly relates to a device and method for modulation and transmission medium for modulating data to data suitable for data transmission and recording in recording media.
2. Description of Related Art
Usually data are modulated to data suitable for transmission and recording when the data are transmitted to a prescribed transmission line or recorded in a recording medium such as an optical disc or a magneto-optic disc. Block coding has been known as one of modulation methods. In the block coding, a data string is converted into blocks composed of units, each unit contains m.times.i bits (referred to as data word hereinafter), and the data word is converted further into code word composed of n.times.i bits according to a suitable code rule. The code word is a fixed-length code if i=1, and the code word is a variable-length code if i is selected from a plurality of numbers, that is, the code word is converted with selection of a prescribed i in a range from 1 to imax (maximum of i). The block coded code is represented as variable-length code (d, k; m, n; r).
The i is referred to as constraint length, and the imax is referred to as r (maximum constraint length). The minimum run d indicates the minimum consecutive number of "0" inserted in a consecutive "1" in a code sequence, and the maximum run k indicates the maximum consecutive number of "0" inserted in a consecutive "1" in a code sequence.
In a compact disc and minidisc (trademark), in the processing of the variable-length code obtained as described herein above, NRZI (Non Return to Zero Inverted) modulation in which "1" indicates inversion and "0" indicates non-inversion is performed, and NRZI-modulated variable-length code (referred to as record waveform string hereinafter) is recorded.
The minimum inversion interval of the record waveform string is represented by Tmin, and the maximum inversion interval is represented by Tmax. The longer minimum inversion interval Tmin namely the larger minimum run d is desirable for higher density recording in linear velocity direction, on the other hand, the shorter maximum inversion interval Tmax namely the smaller maximum run k is desirable in the view point of reproduction of the clock, various methods for modulation have been proposed.
In detail, for example, RLL (2-7) has been known as a method for modulation used for magnetic disks or magneto-optic discs. The parameter of this modulation method is (2, 7; 1, 2; 3), and the bit interval of the record waveform string is represented by T, then the minimum inversion interval Tmin (=(d+1)T) is 3 (=2+1)T. The bit interval of the data string is represented by Tdata, then the minimum inversion interval Tmin is 1.5 (=(m/n).times.Tmin=(1/2).times.3)Tdata. The maximum inversion interval Tmax (=(k+1)T) is 8 (=7+1)T(=(m/n).times.Tmax)Tdata=(1/2).times.8 Tdata=4.0 Tdata. The detection window width Tw (=(m/n)T) is 0.5 (=1/2)Tdata.
Alternatively, RLL (1-7) has been known as a modulation method used also for magnetic disks or magneto-optic discs. The parameter of this modulation method is (1, 7; 2, 3; 2), the minimum inversion interval Tmin is 2 (=1+1)T (=(2/3).times.2 Tdata=1.33 Tdata). The maximum inversion interval Tmax is 8 (=7+1) T (=2/3).times.8 Tdata=5.33 Tdata). The detection window width Tw is 0.67 (=2/3)Tdata.
In comparison between RLL (2-7) and R (1-7), RLL (2-7) namely the minimum inversion interval Tmin of 1.5 Tdata is more desirable than RLL (1-7) namely 1.33 Tdata for higher density recording in the linear velocity direction, for example, in a magnetic disk system and magneto-optic disc system. However, RLL (1-7) which is said to have larger detection window width Tw and larger jitter tolerance than RLL (2-7) is actually used more popularly.
An example of RLL (1-7) conversion table is shown herein under.
TABLE 1 ______________________________________ RLL (1, 7; 2, 3; 2) data code ______________________________________ i = 1 11 00x 10 010 01 10x i = 2 0011 000 00x 0010 000 010 0001 100 00x 0000 100 010 ______________________________________
x in the conversion table is converted to 1 if the subsequent channel bit is 0, or converted to 0 if the subsequent channel bit is 1 (the same is true hereinafter). The constraint length r is 2.
This code can be realized by inverting the order of each bit ranging from MSB to LSB as listed in the table shown herein under.
TABLE 2 ______________________________________ RLL (1, 7; 2, 3; 2) data code ______________________________________ i = 1 11 x00 10 010 01 x01 i = 2 0011 x00 000 0010 010 000 0001 x00 001 0000 010 001 ______________________________________
The constraint length r is 2.
Alternatively, the RLL (1-7) conversion table can be realized also by changing an array as described herein under.
TABLE 3 ______________________________________ RLL (1, 7; 2, 3; 2) data code ______________________________________ i = 1 00 00x 01 010 10 10x i = 2 1100 000 010 1101 000 00x 1110 100 010 1111 100 00x ______________________________________
The constraint length r is 2.
This code can be realized by inverting the order of each bit ranging from MSB to LSB as listed in the table shown herein under.
TABLE 4 ______________________________________ RLL (1, 7; 2, 3; 2) data code ______________________________________ i = 1 00 x00 01 010 10 x01 i = 2 1100 010 000 1101 x00 000 1110 010 001 1111 x00 001 ______________________________________
The constraint length r is 2.
Further, an example of the conversion table of RLL (1-6) having a reduced maximum inversion interval Tmax at the minimum run d=1 is shown herein under.
TABLE 5 ______________________________________ RLL (1, 6; 2, 3; 4) data code ______________________________________ i = 1 11 10x 10 010 01 00x i = 2 0011 100 010 0010 100 00x 0001 000 010 i = 3 000011 000 001 010 000010 000 001 00x 000001 100 000 010 i = 4 00000011 000 001 000 010 00000010 000 001 000 00x 00000001 101 000 000 10x 00000000 001 000 000 10x ______________________________________
The constraint length r is 4.
This code is realized by inverting the order of each bit ranging from MSB to LSB as listed in the table shown herein under.
TABLE 6 ______________________________________ RLL (1, 6; 2, 3; 4) data code ______________________________________ i = 1 11 x01 10 010 01 x00 i = 2 0011 010 001 0010 x00 001 0001 010 000 i = 3 000011 010 100 000 000010 x00 100 000 000001 010 000 001 i = 4 00000011 010 000 100 000 00000010 x00 000 100 000 00000001 x01 000 000 101 00000000 x01 000 000 100 ______________________________________
The constraint length r is 4.
In addition to the above-mentioned cases, an example conversion table of RLL (1-6) code having a reduced maximum inversion interval Tmax and worst error transmission of finite pit is listed in the table shown herein under.
TABLE 7 ______________________________________ RLL (1, 6; 2, 3; 5) data code ______________________________________ i = 1 11 10x 10 010 01 00x i = 2 0011 100 010 0010 100 00x 0001 000 010 i = 3 000011 000 001 010 000010 000 001 00x 000001 100 000 010 i = 4 00000011 000 001 000 010 00000010 101 000 000 10x 00000001 001 000 000 10x i = 5 0000000011 101 000 000 100 010 0000000010 101 000 000 100 00x 0000000001 001 000 000 100 010 0000000000 001 000 000 100 00x ______________________________________
The constraint length r is 5.
Table 7 is arranged so that error does not continue infinitely when pit shift error happens to occur and demodulated. Infinitely continuing error means that when once a pit error happens to occur at a certain point in a code string all the following data strings are not demodulated correctly.
The pit shift error means the error arising from shifting of "1" for indicating an edge forward or backward in a code string generated by modulating a data string.
In addition to the above-mentioned tables, an example of the conversion table of RLL (2-7) having the maximum inversion interval Tmax of 8T (maximum run of 7) at the minimum ran d=2 is listed in the table shown herein under.
TABLE 8 ______________________________________ RLL (2, 7; 1, 2; 3) data code ______________________________________ i = 1 11 1000 10 0100 i = 2 011 001000 010 100100 000 000100 i = 3 0011 00001000 0010 00100100 ______________________________________
Occurrence frequency of the channel bit string subjected to RLL (1-7) modulation is mostly Tmin namely 2T, and followed by 3T and subsequently by 4T. The case that the edge information occurs very often with a short period such as 2T or 3T is advantageous for clock reproduction, however, consecutive 2T results in distorted recording waveform because the waveform output of 2T is small and susceptible to defocus and tangential tilt. A record having consecutive minimum marks is susceptible to external disturbance such as noise in high linear density condition, and erroneous data reproduction occurs often.
RLL (1-7) is combined often with PRML (Partial Response Maximum Likelihood) to improve S/N in reproduction of high linear near density record. The method is used for Viterbi decoding of RF reproduction waveform which is equalized to, for example, PR (1, 1) or PR (1, 2, 1) to match the media characteristics. For example, the reproduction output which is desirable for equalization to PR (1, 1) is listed herein under.
______________________________________ 1 0 1 0 1 0 1 0 (channel bit data string) 1 1 0 0 1 1 0 0 (after NRZI conversion) . . . 1 1 1 1 -1 -1 -1 -1 1 1 1 1 1 -1 . . . . . . +2 0 -2 0 +2 0 . . . (reproduction output) ______________________________________
The data after NRZI conversion is a level data, the value is a value which is different (0 or 1) from the immediately antecedent value (1 or 0) if a channel bit data is 1, on the other hand, the value (0 or 1) is the same as the immediately antecedent value (0 or 1). In this example, "11" is decoded if the value after NRZI conversion is 1, on the other hand, "-1-1" is decoded if the value after NRZI conversion is 0. When Tmin namely 2T is consecutive, the waveform is equalized so that this reproduction output is outputted. Generally, the higher is the liner density the longer the waveform interference is, and the waveform equalization becomes long like PR (1, 2, 2, 1) or PR (1, 1, 1, 1).
When the suitable waveform equalization is PR (1, 1, 1, 1) as the result of application of high linear density at the minimum run d=1, in view of consecutive Tmin namely 2T, the reproduction signal is shown herein under.
______________________________________ 1 0 1 0 1 0 1 0 1 0 (channel bit data string) 1 1 0 0 1 1 0 0 1 1 (after NRZI conversion) . . . 1 1 1 1 1 1 1 1 -1 -1 -1 -1 -1 -1 -1 -1 1 1 1 1 1 1 1 1 1 -1 -1 -1 . . . . . . 0 0 0 0 . . . (reproduction output) ______________________________________
0 continues consecutively. This reproduction output indicates that no signal is outputted continuously after waveform equalization, therefore Viterbi decoding does not merge, therefore this situation results in unstable data reproduction and clock reproduction.
For example, when a data string before modulation is "10-01-10-01-10- . . . " of RLL (1, 7; 2, 3; 2) in Table 1, such channel bit data string is outputted.
A data string before modulation of "11-10-11-10-11-10- . . . " of RLL (1, 6; 2, 3; 4) in Table 5 results in the same.
A data string before modulation of "010-010-010-010- . . . " of RLL (2, 7; 1, 2; 3) in Table 8 results in the same.
In the case that high density recording is applied to a recording media such as magnetic disk, magneto-optic disc, or optical disc, when a code having a long minimum run such as RLL (1-7), RLL (1-6), RLL (2-7), or VFM code as a modulation code, reproduction output of Tmin of the record reproduction waveform at high liner density becomes small. Therefore, if the minimum inversion interval Tmin occurs consecutively, generation of a clock is affected adversely by the occurrence and by the external disturbance such as jitter, and errors is apt to occur, such problem is disadvantageous.
Similarly, in the case that PR (1, 1, 1, 1) equalization is performed using d=1 code under high linear density condition, a signal of consecutive 0 is outputted as a reproduction signal, Viterbi decode does not merge, and clock reproduction is affected adversely.